Perhaps, to make this initially more UPS-friendly, a refill amount incrementer, like 100, could be used: refill_amount+100, instead of refill_amount+1

In the mod's current state, I imagine initial performance costs can be steep with really low patch amounts (1, 2, 3, 4....). Yes, it gets better in the long run, but allowing a higher refill incrementer lessens the impact. I don't know. I barely play this game. Just like reading how people make their mods.

And you might need to, if the above is implemented, set a cap on storage.refill_amount, juuust in case it gets to the ResourceEntity's "amount" limit (uint32 or around 4 billion).

Hmm... I just realized something... on_resource_depleted is called for every tile, and this refill amount is incremented for every tile depleted. So, that amount increases quicker than what I thought. Again, I don't play factorio much.

And this means that newly mined resource patches will get potentially much larger refill amounts than the patches before it.

Yeah, based on my testing this is not a real issue, the game will quickly get to over 1000 and then it stops becoming a real issue. I've found that the UPS impact is always very minimal unless you have stuff running every tick, which I don't. To your second point, someone would have to mine 8 * 10^18 ore for integer overflow to become an issue, which just isn't happening. Even and 1,000,000 ore mined per second (already unreasonable), it would take a quarter of a million years to deplete enough resources for an overflow to happen.